Required Reading

Alex Tabarrok. 

For the record, I partially disagree with this approach as I think we can go a long way under the assumption of homogeneous response and looking at average causal effects of medication treatment.  In a lot of ways this simplifies what could be a confusing nightmare to regulate.  It's a classic balance of harms: omitting some medications (false negatives) to reduce the number of drugs that are actually harmful (false positives).  Without infinite power you need to draw a line somewhere.

But if you are a defender of the traditional FDA it is actually imperative that you become familiar with the argument from personalized medication.  After all, it is the approach that is going to find all of your errors and may well lead to the updating of the current consensus.